Eddy mean flow decomposition and eddy-diffusivity estimates in the tropical Pacific Ocean 1. Methodology
Eddy mean flow decomposition and eddy-diffusivity estimates in the tropical Pacific Ocean 1. Methodology
复制标题
DOI：
10.1029/1998jc900009
复制
发表时间：
1998-12-15
影响因子：
3.6
通讯作者：
Owens, K
中科院分区：
文献类型：
--
作者：
Bauer, S;Swenson, MS;Owens, K
由于赤道和近赤道动力学的快速响应时间，热带太平洋表层流系统具有很强的非平稳性。海洋-大气动力学在大尺度上产生具有较强经向切变(L(Y)类似纬度1度)的纵向相干纬向气流(纬向长度尺度L(X)类似于60度)和一个高能中尺度(O(100公里))分量。中尺度场效应的参数化取决于大尺度平均速度与观测速度的分离。本文的重点是：将流动分离为大尺度、平均和中尺度涡旋分量，以便在表现出强气流和强切变的流区计算有意义的涡扩散系数。大尺度平均中的大梯度排除了传统的入库技术估计扩散率的可能性。在这两篇出版物中的第一篇中，发展了一种使用拉格朗日数据来估计扩散系数的方法，以解决平均流的不均匀问题。用最小二乘双三次光滑样条插值方法计算平均光场的空间相关估计，并通过优化的粗糙度参数来保证低频起伏光场中的能量最小。基于湍流剪切流随机模型的数值模拟被用来在一个概念上简单但现实的场景中验证我们的方法。这项技术被应用于1979-1946年在热带太平洋两个动态不同的时空区域获得的近地表漂流观测。在南赤道流中，第一个区域的特征是线性纬向切变平均流，残差具有近似指数的自协方差结构。两个分量的速度残差均为(S)=130 cm(2)S(-2)，水平扩散系数(U)约为7×10(7)cm(2)S(-1)，(V)约3×10(7)cm(2)S(-1)。估计值没有明显的年际变化，但估计值中的残差趋势是由3个月季节期间速度场的季节内变化引起的。第二个区域位于北赤道逆流和北赤道流之间，为纬向切变较强、向北速度较弱的平均气流。纬向分量的自协方差近似为指数，而经向分量在10天左右有一个负叶，这可能是由于不稳定波的存在。纬向分量的方差为380 cm(2)，经向分量的方差为360 cm(2)，S(-2)的方差为360 cm(-2)，水平扩散系数(U)约为15×10~(7)cm(-1)，S(-1)和S(-1)的水平扩散系数约为4×10(7)cm(-1)。强烈的季节内变异性要求最多2个月的时间窗口，以便在协方差计算中保持近似平稳。
The tropical Pacific Ocean surface current system can be characterized by a strong degree of nonstationarity due to the fast response time of equatorial and near-equatorial dynamics. The ocean-atmospheric dynamics create longitudinally coherent zonal flow (zonal length scales l(x) similar to 60 degrees) with strong meridional shear (l(y) similar to 1 degrees in latitude) in the large-scale mean and an energetic mesoscale (O(100 km)) component. Parameterization of the effects of the mesoscale field depends on the separation of the large-scale mean from the observed velocity. In this paper the focus is placed on the key issue: separating the flow into large-scale mean and mesoscale eddy components in order to compute meaningful eddy diffusivity estimates in flow regimes that demonstrate strong currents and strong shear. Large gradients in the large-scale mean have precluded diffusivity estimation by traditional binning techniques. In this first of two publications, a method is developed for using Lagrangian data to estimate the diffusivity addressing the inhomogeneity of the mean flow. The spatially dependent estimate of the mean field is computed with a least squares bicubic smoothing spline interpolation scheme with an optimized roughness parameter which guarantees minimum energy in the fluctuation field at low frequencies. Numerical simulations based on a stochastic model of a turbulent shear flow are used to validate our approach in a conceptually simple but realistic scenario. The technique is applied to near-surface drifter observations obtained from 1979-1946 from two dynamically distinct time-space regions of the tropical Pacific Ocean. The first region, in the South Equatorial Current, is characterized by a linear zonal shear mean flow and an approximately exponential autocovariance structure in the residuals. The velocity residuals have velocity variance of (s) over cap(2) = 130 cm(2) s(-2) for both components, and horizontal diffusivities are (u) approximate to 7 X 10(7) cm(2) s(-1) and (v) approximate to 3 X 10(7) cm(2) s(-1). No significant interannual variations of the estimates are detected, but residual trends in the estimators arise from intraseasonal variations in the velocity field during the 3-month season. The second region, in the North Equatorial Countercurrent and the North Equatorial Current, has a mean flow with a strong zonal shear and a weak northward velocity. The autocovariance is approximately exponential for the zonal component, while the meridional component has a negative lobe at about 10 days, probably due to the presence of instability waves. The variance is 380 cm(2) s(-2) for the zonal component and 360 cm(2) s(-2) for the meridional component, while the horizontal diffusivities are (u) approximate to 15 X 10(7) cm(2) s(-1) and (v) approximate to 4 X 10(7) cm(2) s(-1). Strong intraseasonal variability requires a maximum time window of 2 months for approximate stationarity to hold for the covariance calculations.